chore(main): release leantoken 0.2.0

[github-actions]

Changelog

All notable changes to this project will be documented in this file.

[0.2.0] - 2026-10-17

Benchmarks

• indexing: Record exact token-count reuse rejection (bench(indexing): record exact token-count reuse rejection #629) (#629)

Bug Fixes

• release: Preserve API compatibility and validate candidates (fix(release): preserve API compatibility and validate release candidates #661) (#661)

• test: Align cold MCP initialization with doctor deadline (#660)

• setup: Identify and retain private runtime artifacts (fix(setup): address private runtimes by full artifact digest #654) (#654)

• ci: Normalize source roots for coverage inventory (fix(ci): normalize source roots for coverage inventory #659) (#659)

• ci: Classify reported test modules with the source inventory (fix(ci): classify reported test modules with the source inventory #658) (#658)

• cache: Bound inventory work independently of pagination (fix(cache): bound inventory independently of pagination #653) (#653)

• BREAKING read: Distinguish unknown target ends from scan frontiers (fix(read)!: distinguish unknown target ends from scan frontiers #651) (#651)
  Migration: ReadResponse.target_end_line is now Option, serialized as null for an open-ended bounded target whose end is unknown. observed_end_line reports the scan frontier separately.

• bench: Enforce MCP readiness response deadline (fix(bench): enforce MCP readiness response deadline #647) (#647)

• bench: Reject nonfinite regression thresholds (fix(bench): reject nonfinite regression thresholds #646) (#646)

• mcp: Start indexing for stateless sessions (fix(mcp): start indexing for stateless sessions #649) (#649)

• npm: Forward signals while the native child is running (fix(npm): forward signals while the native child is running #645) (#645)

• cli: Reject malformed batched history targets

• benchmarks: Validate files continuations before comparing identities (fix(benchmarks): validate files continuations before comparing identities #638) (#638)

• benchmarks: Reject failed MCP profiler shutdown (fix(benchmarks): reject failed MCP profiler shutdown #637) (#637)

• bench: Separate MCP takeover acquisition and publication (fix(bench): separate MCP takeover acquisition and publication #630) (#630)

• benchmarks: Normalize matching native receipt identities (fix(benchmarks): normalize matching native receipt identities #626) (#626)

• profiling: Parse ANSI-colored watcher diagnostics (fix(profiling): parse ANSI-colored watcher diagnostics #625) (#625)

• mcp: Keep ready receipt hits from activating dormant contexts (fix(mcp): keep ready receipt hits from activating dormant contexts #624) (#624)

• indexing: Honor watcher backpressure retry deadlines (fix(indexing): honor watcher backpressure retry deadlines #621) (#621)

• ci: Exclude stale workspace coverage maps (fix(ci): clean cached workspace artifacts before coverage #620) (#620)

• ci: Parse NUL-delimited Git changed paths (fix(ci): parse NUL-delimited Git changed paths #619) (#619)

• mcp: Preserve direct replies across receive cancellation (fix(mcp): preserve direct replies across receive cancellation #614) (#614)

• testing: Align fixtures, validation, coverage and release profiling (fix(testing): align fixtures, validation, coverage and release profiling #601) (#601)

• core: Preserve bounded evidence and runtime ownership (fix(core): preserve bounded evidence and runtime ownership #600) (#600)

Chores

• test: Bound local concurrency and streamline focused runs

Documentation

• release: Distinguish candidate validation from PR approval (docs(release): distinguish candidate validation from PR approval #663) (#663)
• measurement: Preserve rejected tokenizer ownership screens (docs(measurement): preserve rejected tokenizer ownership screens #640) (#640)
• Publish shared-cache MCP memory characterization (docs: publish shared-cache MCP memory characterization #639) (#639)
• measurement: Characterize large-tree reconciliation scope (docs(measurement): characterize large-tree reconciliation scope #633) (#633)
• mcp: Clarify argument error recovery boundary (docs(mcp): clarify argument error recovery boundary #631) (#631)
• context: Distinguish receipt suppression from known hashes (docs(context): distinguish receipt suppression from known hashes #628) (#628)
• context: Clarify soft focus coverage minimums (docs(context): clarify soft focus coverage minimums #627) (#627)

Features

• cache: Compact selected inactive indexes (feat(cache): compact selected inactive indexes #642) (#642)
• benchmarks: Report proportional and private MCP memory (feat(benchmarks): report proportional and private MCP memory #634) (#634)

Refactoring

• test: Consolidate process and frozen-report coverage
• setup: Reuse launcher path matcher
• mcp: Share resource protocol policy
• mcp: Centralize legacy protocol shape
• cursors: Remove unused legacy version paths

Testing

• doctor: Keep catalog fixtures valid across version bumps (test(doctor): keep catalog fixtures valid across version bumps #662) (#662)
• storage: Upgrade authentic historical migration fixtures (test(storage): verify authentic historical SQLite upgrades #655) (#655)
• cache: Validate inventory age within each inspection (test(cache): validate inventory age within each inspection #657) (#657)
• storage: Bind query-plan evidence to production SQL (test(storage): bind query-plan evidence to production SQL #656) (#656)
• ci: Classify indexed service fixtures as cold SQLite (test(ci): classify indexed service fixtures as cold SQLite #652) (#652)
• mcp: Capture receipt rebase timeout phases (test(mcp): capture receipt rebase timeout phases #650) (#650)
• npm: Exercise the installed product and simplify packaging
• Prune redundant units and repair flaky assertions
• ci: Fix Windows rollback and diff fixtures (test(ci): fix Windows rollback and diff fixtures #622) (#622)
• mcp: Use shared protocol shape policy
• Remove retired wire compatibility checks
• Drop unused changed-path compatibility mode
• Remove retired compatibility assertions

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-17T06:06:08.921204Z	4e4ccb2	Manual request

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 3d61abafd2

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".

[N0zoM1z0]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Can't wait for the next one!

Reviewed commit: 303a32a20d

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[N0zoM1z0]

@codex review

Please review the regenerated candidate 8049d04, including the version-independent doctor fixture from #662. The generator automatically dispatched CI for this exact head; this remains a release candidate pending validation.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Keep it up!

Reviewed commit: 8049d04ab0

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[N0zoM1z0]

Validated regenerated candidate 8049d04ab0d600d808d386ebb0844c3d39b795c2 after #661 and #662.

• All four package identities are 0.2.0; the generated PR body exactly matches the latest changelog section, including the breaking ReadResponse.target_end_line migration.
• The generator explicitly dispatched complete CI for this exact candidate. All 15 planned evidence jobs and Required checks passed. Downloaded receipts match the exact source, lane, runner, command, topology, schema and planner identities.
• Actual 0.2.0 product logs show 1088 Linux, 1084 macOS and 1020 Windows tests passed with zero skips. Coverage, repeated test profiling, Rust quality, npm installation, dependencies, examples/docs and release planning also passed. Release planning ran all five real Git/git-cliff fixture tests and announced v0.2.0.
• Codex and Greptile reviewed this exact head without new findings; prior release review threads are resolved. Post-merge main CI also passed.

The earlier candidate failure was reproduced rather than retried away: its version-hardcoded catalog fixture failed on all three platforms/coverage/profiling, and Required checks rejected its failed receipt. #662 corrects that fixture while preserving current-release catalog strictness. This PR remains open as a validated release candidate; no tag or package was published by this follow-up.

[N0zoM1z0]

@codex review

Regenerated candidate 4e4ccb2 includes #663. Its only differences from fully validated 8049d04 are release documentation and the generated changelog. Four 0.2.0 identities, body/changelog equality and the breaking migration were rechecked. Both automatically dispatched complete CI and the approved original PR workflow will run for this current candidate.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. What shall we delve into next?

Reviewed commit: 4e4ccb272a

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[N0zoM1z0]

Final validation after #663 for candidate 4e4ccb272a91604c3f0abcd4edde29552c6b1c0c:

• Complete dispatched candidate CI: all 15 planned evidence jobs and Required checks passed, including three-platform products/contracts, Rust quality, coverage, repeated profiling, dependencies, examples/docs, npm installation and release planning.
• Approved original PR CI: all six selected evidence jobs and Required checks passed. The PR check summary now contains the actual checks and GitHub reports CLEAN.
• Downloaded receipts were independently compared with each immutable plan. The dispatched plan/15 receipts bind to the exact candidate head; the PR plan binds to that head and its GitHub merge revision 78c0692601734e50266d6e486fcaa35b0bd287cb, with all six receipts matching. Source, lane, runner, command, topology, schema and planner identities agree.
• Latest main CI passed; its 11/11 planned receipts bind to 5cfab83047d4dc77b1b5826016bc75edcd9b4bc4.
• Rechecked four 0.2.0 package identities, exact changelog/PR-body equality and the breaking ReadResponse migration. Codex and Greptile reviewed this current candidate without new findings. No unresolved release review threads remain.

This remains an open, validated release candidate. No tag or package was published by this follow-up.